How can we improve the way we measure poverty? The UN's new poverty index (and groovy graphics)

July 28, 2010

The co-creator of the UN's new Multidimensional Poverty Index defends her new baby

July 28, 2010

Guest Blog: World Bank research director critiques the new UN poverty index

July 28, 2010
empty image
empty image

Martin Ravallion is Director of the World Bank’s research department, the Development Research Group. These are martin ravallionthe views of the author, and need not reflect those of the World Bank.

“Everyone agrees that poverty is not just about low consumption of market commodities by a household.  There are also important non-market goods, such as access to public services, and there are issues of distribution within the household. It is agreed that consumption or income poverty measures need to be supplemented by other measures to get a complete picture.

But does that mean we should add up the multiple dimensions of poverty into a single composite index? Or should we instead measure consumption poverty with the best data available, while also looking for the best data on other dimensions of poverty as appropriate to the country context?

The Oxford Poverty and Human Development Initiative (OPHI) has recently launched a Multidimensional Poverty Index (MPI), and calculated it for over 100 countries.   The MPI is a composite of indicators selected for consistency with the UNDP’s famous Human Development Index (HDI). The HDI uses aggregate country-level data, while the MPI uses household-level data, which is then aggregated to country level. The index has ten components; two represent health (malnutrition, and child mortality), two are educational achievements (years of schooling and school enrolment), and six aim to capture “living standards” (including both access to services and proxies for household wealth).  The three broad categories–health, education, and living standards–are weighted equally (one-third each) to form the composite index.   

One can debate the precise indicators chosen for the MPI by the Oxford team (who are clearly aware of the many data concerns). For example, the MPI’s six “living standard” indicators are likely to be correlated with consumption or income, but they are unlikely to be very responsive to economic fluctuations.  The MPI would probably not capture well the impacts on poor people of economic downturns (such as the Global Financial Crisis) or rapid upswings in macro-economic performance.

The precise indicators used in the MPI were not in fact chosen because they are the best available data on each dimension of poverty. Rather they were chosen because the methodology used by the MPI requires that the analyst has all the indicators for exactly the same sampled household. So they must all come from one survey. There is much better data available on virtually all of the components of the MPI, but these better data can’t be used in the MPI since they are only available from different surveys. This aspect of their methodology greatly constrains the exercise. If one chooses not to form the composite at household level but to look instead at the separate dimensions of poverty one is free to choose the best available data on each dimension of poverty.

There is a deeper concern about the MPI, which holds even if the best data all came from just one survey. The index is essentially adding up “apples and oranges” without knowing their relative price. When one measures aggregate consumption from household-survey data for the purpose of measuring poverty, as in the World Bank’s “$1 a day” measures, one relies on economic theory, which says that (under certain conditions) market prices provide the correct weights for aggregation. We have no such theory for an index like the MPI. A decision has to be taken, and no consensus exists on how the multiple dimensions should be weighted to form the composite index. 

On closer scrutiny, the embedded trade-offs (stemming from the weights chosen by the analyst) can be questioned, and may be unacceptable to many people.  In the context of the HDI, I pointed out 15 years ago that by aggregating GDP per capita with life expectancy the HDI implicitly put a value on an extra year of life, and I showed that this value rises from a very low level in poor countries to a remarkably high level in rich ones (4-5 times GDP per capita).   If it was made clearer to users, I expect that they would question this trade-off embedded in the HDI.

The MPI index faces the same problem. How can one contend (as the MPI does implicitly) that the death of a child is equivalent to having a dirt floor, cooking with wood, and not having a radio, TV, telephone, bike or car?  Or that attaining these material conditions is equivalent to an extra year of schooling (such that someone has at least 5 years) or to not having any malnourished family member?  These are highly questionable value judgments. Sometimes such judgments are needed in policy making at country level, but we would not want to have them buried in some aggregate index.  Rather, they should be brought out explicitly in the specific country and policy context, which will determine what trade off is considered appropriate; any given dimension of poverty will have higher priority in some countries and for some policy problems than others. 

Poverty is indeed multidimensional.  But it is not obvious how a composite multidimensional poverty index such as the MPI contributes to better thinking about poverty, or better policies for fighting poverty.  Being multidimensional about poverty is not about adding up fundamentally different things in arbitrary ways. Rather it is about explicitly recognizing that there are important aspects of welfare that cannot be captured in a single index.”

Sabina Alkire of OPHI (and the creator of the MPI) responds tomorrow. For Duncan’s introductory post on the MPI see here.


  1. I would like to query the World Bank approach to measuring poverty. Such a complex phenomenon cannot be reduced to a mere measure of consumption. The development of a composite index, whatever its flaws, is preferable to this simplistic approach. A composite index, by its very nature, includes multiple aspects of life. This is preferable to covering only ONE highly suspect issue – consumption, taking purchasing power parity (if this exists) into account. The so-called economic theory, which says that “(under certain conditions) market prices provide the correct weights for aggregation” is highly flawed. This reflects neo-liberasl ideology rather than proper economic theory. In measuring consumption, the attempts of the World Bank to ascribe a value to a shack where people live, or to home grown produce, comparable throughout the world, are laughable. The GDP itself is a highly questionable index. If I am involved in a car crash, and I require hospitalisation, this contributes to GDP! What about quality of life? So we should all reject this World Bank attempt at measuring poverty and send them back to the drawing board. It tells us nothing. Instead, we should favour any measurement which attempts to understand a complex phenomenon through considering a wider range of issues. While there will always be questions regarding the accuracy and relevance of the calculation of an index, surely the debate that this encourages is preferable to promoting World Bank and IMF dogma.

  2. And, of course, there is the critique (that can equally be leveled at the MPI and other more parsimonious measures or poverty) that many of these debates over how to accurately measure poverty are little more than a distraction from examining the structures that produce material inequality. We do need information regarding who the poor are, how they live, and where they live. But this alone is only a beginning, and knowing ever more precisely just who and where the poor are is a poor substitute for change.

  3. Rosaline:

    Ravallion *is* saying we need more than one approach to measure poverty:

    “Poverty is indeed multidimensional. But it is not obvious how a composite multidimensional poverty index such as the MPI contributes to better thinking about poverty, or better policies for fighting poverty. Being multidimensional about poverty is not about adding up fundamentally different things in arbitrary ways. Rather it is about explicitly recognizing that there are important aspects of welfare that cannot be captured in a single index.”

    Also, your GDP interpretation is a fallacy – imagine you are on your way to buy a $300 tv when you get in a car accident. You are uninsured and so you have to fork out for your $600 hospital bill yourself, so you must cut your own spending by $600 and use that cash for the hospital bill. Your spending has decreased elsewhere (including the telision), so the total change in GDP is 0.

    You can get a more detailed explanation here:

  4. It is a generalized problem in development to aggregate all kind of issues into one big, undefined drab.

    Indeed, while air pollution is real and measurable, ‘sustainable development’ mains essentially nothing, and is in the end a synonym for ‘multi-dimensional poverty’. Choices must be made between apples and oranges, child protection and child survival, implicitly or explicitly. Indeed, there are interactions: but apart from win-wins, there are also the win-lose or even lose-lose interactions. This is why these choices are made in a democracy during a democratic budgeting process, not in a technocratic planning exercise.

  5. Ravallion falsely implies that there are value judgments in the MPI but there are not value judgments in the Bank’s IPL. The Bank’s IPL is also value laden.
    Ravallion then contends that we should look for the best data in each dimension of deprivation, rather than from a single survey source. But this results in a highly problematic outcome. We cannot know whether one individual has suffered two deprivations or two people have each suffered one. This matters morally, as the former case is worse, and for policy, as addressing a group of deprivations will be different than a single deprivation.
    Finally, his critique of the difficulty of providing relative weights to different deprivations is very weak. Just because it is difficult to compare health indicators to education indicators does not mean that we should solve the problem by running away from it. Rather, we meet it head on and have a public discussion about the difficult value-laden choices that have to be made in evaluating deprivation. Public policy must make such choices all the time.
    Ravallion concludes with nonsense:” But it is not obvious how a composite multidimensional poverty index such as the MPI contributes to better thinking about poverty, or better policies for fighting poverty. Being multidimensional about poverty is not about adding up fundamentally different things in arbitrary ways. Rather it is about explicitly recognizing that there are important aspects of welfare that cannot be captured in a single index”. The Alkire Santos paper clearly highlights how it does matter. It shows policy makers the various deprivations that people face, and paints a very different picture of the distribution of poverty. That matters for poverty alleviation. Endorsing a single index does not mean that one doesn’t recognize that there are important aspects of welfare that are incommensurable.

  6. Very interesting debate indeed. Although my knowldge is modest in this field, i don’t agree with Ravaillon on two things: 1/ use different source of datainstead of one survey would be better. we all know that wehn using different source of data we can’t measure anything (other than individual figures coming from different sources) and more importantlywan can’t infere the findings and analysis to the wider population. 2/ consumption based measures are based on economic theory while deprivation measures are not. there is extensive litterature and knwoledge produced in the last few decades on both deprivation, multidmenionality of poverty and measurements issues where you can see critics, self critics and self awareness of limitations of developing and using indexes. the most important things as Alkire pointed out in her reply is that the methodlogy should be transparent so that we interpret those findings within the limitations of the index. this is the only way those measurements would improve to provide a better picture of poverty and provide more effective answers. we’ll not wait until we have a perftect measure with no weakness, that seems impossible anyway !

  7. Todos consensuamos que la forma de medicion de la pobreza planteada es muy simple y que requiere adicionar nuevos indicadores como la diferenciacion de zonas rurales, urbanas, urbano marginales.
    además en varios paises los estados utilizan estrategias para reducir la pobreza como la entrega de dinero, alimentos. que en lo inmediato es bueno pero no genera sostenibilidad en el futuro, creo que se debe enseñar a pescar y no solamente a comerselo

    Duncan: thanks Henry, here’s a rough translation for non Spanish speakers: “We all agree that the usual way to measure poverty is very simple and that new indicators should be added such as differentiation between rural, urban and marginal urban zones. Also in several countries, states use various strategies to reduce poverty, such as cash or food transfers, which is good in the short term, but does not generate long term sustainability. I think you need to people to fish, not just give them fish to eat.”

  8. This MPI measurement is noble one. This will capture real prevalent of poverty population in any society. It will also maintain and measure the real chronic poor, poor and non-poor of any country by adopting not only income or xpenditure but also incorporate with many dimensions related to poverty such are, health, social and economic dimensions of poverty.

  9. I am pleased to read the various views on mesuring poverty and their implications to policy and decision making.

    Despite the challenges, I believe, sufficent and accurate date collected at the same time (season)for all the indicators that would be included in to the MPI equation is the cornerstone. Without this, we end up in wrong conclusion due to variations caused by the methodolgy used, seasonality of the data as well as the tools used for data analysis.

    Furthermore, the MPI provides all options i.e., if we want to see one aspect of the componenets poverty for instance income, or health poverty or the combined/aggregate aspect of poverty it has it.

    For me the policy development and subsequent decision making is the most important issue. A singluar analysis ends to a singular conclusion leading to singular decision. For instance if it is about income poverty, the implication is to make a policy and decide on measures for raising income of the targeted people without looking into the inter and intra houshold or regional distribution. But if we take the composite approach as parameter we will be in a better position to develop a comperhensive policy that targets poverty in a comperhensive manner regardless of the target area or household or an individual. We are lucky though the MPI has the tools to do all these levels i.e., area, household and individual.

    My conclusion is this, with the above conditions (quality and timeliness of data with accurate methodolgy and analysis used), MPI is by far better than the WB singular approach to measure poverty.

  10. Ravallion is advocating some sort of methodological purity in his criticism of MPI while his own methodology is riddled with theoretical and statistical assumptions, and data caveats. MPI is a composite index and its true value would emerge over time, perhaps after at least five years of consistent measurement and evaluation. Let’s face it, Consumer Confidence Index in the US is not the most scientific concept rooted in any deep economic theory, yet by measuring it in a standardized way for years, it has developed some reasonable interpretation and turned into a useful policy indicator. Ravallion should have enough scholarly mettle see the inconsistency in his criticism of MPI. MPI, for all its shortcomings, would be a useful metric to publish on a regular basis.

  11. Well, one thing is perfectly clear – poverty is multidimensional and goes beyond income (expenditure/consumption).
    How best can we measure this multidimensional phenomenon is the biggest question that all development researchers and practitioners have been bogged down in?
    I strongly believe that a dollar a day ($1.25 a day) cannot capture the extent of poverty one is going through. The worst part is governments are rejoicing for reducing poverty by gauging their achievement against this measurement – instead of going back to the drawing board and critically looking at their strategy. The WB– using their financial muscle as an advocacy – is the first institute to clamp their hands in support of such flawed conclusions.
    In order to critically see the extent and depth of poverty a comprehensive measure is needed. This measure does not have to be perfect, rather has to clearly show the problem vividly.
    We all know the problem of aggregation to measure once (household’s) wellbeing – and if the new tool is based on a household data that is a being leap forward.
    Having said this, one alarming issue that all needs to take into account is the quality of data (sometimes doctored data); and one way of minimizing such challenges is triangulation. Looking at the multidimensional (broader) factors is one additional advantage I could see immediately.

  12. Dear, proponents and opponents of MPI
    Poverty is a phenomenon which creates its base first and than creeps and we called it chronic poverty. Secondly, one must know whether the way we are defining poverty is realistic or not. If it is something physical and or behavioral or state of mind. There are number of issues that we have to keep in mind always when we are talking about poverty and its multidimensional nature. This is for sure that this world in the long run future may be in a position to see the end of absolute poverty but what to do about relative poverty that we can give an end to. Most of the literature on poverty until date is rather focusing on relative poverty. A dollar a day or calorie intake or poverty of opportunity. One must think poverty in its early phases as well as its historical and chronological episodes to learn from it and make it history.

  13. OHPI and Martin Ravallion are both right and wrong in my opinion. A single measure of consumption, as used by the World Bank’s under $1.25/day indicators, does not give a true picture of poverty, and so efforts to assess other dimensions of poverty are important. But OHPI’s approach is far too limited (including for the reasons Ravallion describes). Participatory research demonstrates that poor people are extremely concerned about vulnerability to shocks, violence, discrimination, isolation and other negatives – none of which are revealed in the OPHI work. Poverty must also be seen as relative; having much less income that your peers is a factor too (as even Adam Smith recognized). Hence the MPI statistics are all-too-often counter-intuitive. It just isn’t credible that South Africa and the Palestinian Territories are virtually free of poverty; that Ethiopia has much worse poverty than Central African Republic, Burundi and Sierra Leone; that Nicaragua is poorer than Ghana, Philippines, Uzbekistan or Bolivia; or that the former Soviet Union countries are all relatively free of poverty. If indicators don’t tally with the reality as seen with ones own eyes they aren’t credible. OPHI’s papers on Bhutan, for example, make no reference to the marginalization and poverty of the “Nepalis” (those of Madheshi Nepal origin who have lived for generations in Bhutan but who are mostly denied opportunity and citizenship).

    On the other hand, Ravallion’s confidence in income statistics, and the Living Standards Measurement Surveys (LSMS) on which they are based, is misplaced. This approach has methodological flaws, is subject to serious survey errors, produces data that change little in the face of shocks that clearly have a profound impact on the poor, conceal the prevalence of poverty in wealthy countries and produces results that are often unrealistic. For example, tracking the prevailing rates of malnutrition with LSMS poverty data suggests that in about 5 years time the former will overtake the latter – i.e. there will be millions of people who go to bed hungry, even though they aren’t poor. I started writing a paper on this in 2005 when I was working at the World Bank but then got taken over by work on the tsunami reconstruction in Indonesia.

    A single index of poverty is superficially an attractive proposition, but may be unworkable. When I go to a garage I might want to have a single indication of what is wrong with my car; if I am given a “98% good” rating it sounds encouraging, but if the starter motor doesn’t work the car isn’t any use. It’s more useful to have a single rating on what it will take to fix the car (the repair bill). And that might be more practical in the field of poverty. Even that would have to be a dual rating: an estimate of what it would cost to resolve the priority problems of poverty as identified by the poor (as a proportion of GNP, say) plus an estimate of the political will required to achieve the redistribution required.

  14. Such a disappointing critique from Ravallion… “But it is not obvious how a composite multidimensional poverty index such as the MPI contributes to better thinking about poverty, or better policies for fighting poverty.” is such a nonsensical statement.

    A long last, the WB is losing its monopoly power over the production of global poverty statistics.

    Finally, someone proposes and implements an alternative way, as imperfect as it may be, to measure this thing called global poverty. Of course the statistics suffer from limitations, with the data, its coverage, comparability issues, etc. being the principal mischiefs.

    But how are the WB’s methods any better?

    It’s time to let the UN and Oxford’s researchers to do their job and produce their own set of independent-minded, novel, fresh, refreshing, intriguing, multidimensional global poverty statistics.

    Their publication suggests that market for the production of these figures has finally been opened up to competition. Only good things can come out of this.

  15. Yes, I too agree with Sabina C, it is high time to change and WB losing its monopoly power on poverty measurements. MPI is a good starting point, now the experts should try to improve it. Alkire and Sabina has shown that their MPI is open to change, the dimensions, indicators and weights can be changed. So the researchers can tailored it with their requirements and data avilability

    Individual country analysis may be the best usage of MPI and would lead to better policy making than using WB $ 1 (now 1.25).

  16. Poverty is a state of mind. It’s being positive and enthusiastic can win over the adversities of life indicated through various types of deprivations captured by poverty related indices. Mental depression seems to be the cause and effect of what we name as poverty. Affectionate sympathy, moral boosting and institutional support can effectively help overcome the poor state of mind.Including more variables in poverty indices may sound rationally very good but still may not be able even touch the real cause! Poor is one who does not have a friend from among the non poor. Can’t a rich sincerely befriend a poor and thus help solve the problem of poverty? Will this not be able to reduce the painful feeling of inequality as well?

  17. I agree with Sabina Clarendon and Thusitha Kumara in the sense that the MPI is just a start. It has paved a way to furthers researches will may improve it by tackling some the weaknesses that it may present. All in all when we look at poverty as being multifaceted, I can say with a lot of confidence that Ravallion’s critique does not diminish the value that the MPI approach presents.

Leave a comment